true Kuiper-belt object, Gomes’ work gives a new perspective on these bodies. If the hot Kuiper belt is really a record of theUranus–Neptune planetesimal population,a comparison of the detailed physical properties of the hot and cold objects couldprovide valuable information on the way inwhich the temperature and chemical proper-ties of bodies varied with increasing distancefrom the Sun in primordial times. Asastronomers slowly unlock these, and other,mysteries, we will learn more about the earlylife of our planetary system. ■

A. Morbidelli is at the Observatoire de la Côte d’Azur, 06304 Nice, France.e-mail: morby@obs-nice.frH. F. Levison is at the Southwest Research Institute,1050 Walnut Street, Boulder, Colorado 80302, USA.e-mail: hal@boulder.swri.edu

1. Gomes, R. S. Icarus 161, 404–418 (2003).

2. Brown, M. Astron. J. 121, 2804–2814 (2001).

3. Trujillo, C. A. & Brown, M. E. Astrophys. J. 566, 125–128

(2002).

4. Levison, H. F. & Stern, S. A. Astron. J. 121, 1730–1735 (2001).

5. Fernandez, J. A. & Ip, W. H. Planet. Space Sci. 44, 431–439

(1996).

6. Hahn, J. M. & Malhotra, R. Astron. J. 117, 3041–3053 (1999).

Plastids are organelles found in the cellsof higher plants, the best-known formbeing the chloroplast, the site of photo-

synthesis. They arose when, way back intime, plant ancestors assimilated photosyn-thetic prokaryotes — unicellular organismssuch as bacteria which, in contrast to eukaryotes such as plants (and ourselves),typically lack a membrane-bound nucleus.The creation of functional plastids involvedthe migration of huge numbers of genesfrom the prokaryote genome to the plantnucleus. But is transfer of plastid DNA intothe nucleus of higher plants still occurring?

As they describe on page 72 of this issue1,Timmis and colleagues conclude that it is, ata rate comparable to spontaneous mutationof the nuclear DNA. There are various impli-cations here, not least for plant biotech-nology. One proposed strategy for the futureis to engineer genes for some trait or otherinto chloroplasts, rather than nuclei, because— in principle — those genes will affect the characteristics of the individual plantand will be transmitted to future generationsby the maternal parent, but not ‘escape’ inpollen and spread uncontrollably.

The extent of gene migration during evo-lution becomes clear when we consider thatthe plastids of higher plants today containonly about 120 genes, compared to the3,000–4,000 thought to have been present inthe genome of the ancient photosyntheticprokaryote. Plastid function depends onsome 3,000 nuclear genes, the products ofwhich are transferred to complement thoseof the remaining plastid genes2. We have arelatively good idea about the identity andgene content of the organisms that wereinvolved when the original eukaryotic plantancestor took in the prokaryotic forerunnerof the plastid3. But little is known about the

evolutionary process of gene transfer fromthe plastid to the nucleus.

This is what Timmis and colleagues1 setout to study by designing a smart screen toestimate the rate of plastid-to-nucleus DNAtransfer in tobacco plants. Their systeminvolved measuring the transfer rate to theplant nucleus of a gene (neo) — which confersresistance to the antibiotic kanamycin — that had been engineered into the tobaccoplastid genome (along with a marker gene,aadA, of which more later). The kanamycin-resistance gene was endowed with the various signals, including a promoter sequence,required for eukaryote-type expression in the nucleus. Plastids, in general, have a pro-karyotic-type gene-expression machinery4,so the nuclear kanamycin-resistance gene was not expected to be expressed there (Fig. 1a, overleaf).

Timmis’s group found that transfer of neoto the nucleus took place in a total of 16 heritable events in 250,000 seedlings — thatis, at an incidence of around 621015 — andwas detected by testing for kanamycin resistance in the tobacco plants (Fig. 1b, c).Molecular analyses confirmed that the neogenes, together with flanking plastid DNA ofvariable size, had indeed been incorporatedinto the tobacco nuclear DNA at differentgenomic locations.

Another cellular organelle found ineukaryotes is the energy-generating mito-chondrion. Previous work5 with yeast hasshown that here, too, there is transfer oforganelle DNA to the nucleus, and the rate(about 221015 per cell per generation) iscomparable to that found by Timmis et al. in tobacco. In yeast, the fragments of mito-chondrial DNA were incorporated into thechromosomes by a mechanism known as‘double-strand-break repair’6,7, which is

news and views

NATURE | VOL 422 | 6 MARCH 2003 | www.nature.com/nature 31

Plant biology

Mobile plastid genes Pal Maliga

The direct demonstration that chloroplast DNA can be incorporated intothe nuclear genome of plants, even though it is unlikely that such DNAwould be functional, will influence thinking in plant biotechnology.

100 YEARS AGOAll inquirers have perceived that great men are of two types, and it would conduceto clear thinking if we could accustomourselves to classify them under differentnames… The first class, to which I shouldprefer to restrict the name genius, may bedescribed primarily as men of fine, delicate,sensitive, impressionable constitution, and strong, restless innate tendencies which appear early in life, as a rule, and take their own shape. These men workenergetically, often at high pressure, and in general die comparatively young… Thesecond class I would describe as men oftalent. When preeminent they exhibit striking aptitude in learning and in imitation, and develop extraordinary powers of work… In nature there is greatvariety, and genius, so far, is one of thevarieties which often recur, but scarcely ever survive even for two generations. It is a rare and delicate thing, and the utmostwe can hope for it is that endeavours may be made to collect and preserve it like some hot-house plant, in order that it maysuggest combinations which men of talentmay put to practical account. The position of the second type in the struggle forexistence is beyond doubt. The stability of a country and its place among the nationsdepend upon the number and ability of menof this stamp.From Nature 5 March 1903.

50 YEARS AGOThere has been a progressive increase, both absolute and relative, in the proportionof old people in the population, this increasebeing the result of the decline in infant and young adult mortality produced by themedical and social advances that have beenmade possible by the application of thescientific method during the past century. As a result of this, more old people must besupported by a diminished proportion ofwage earners. There is a correspondingincrease in the numbers of elderly invalids to be cared for, who are suffering from the degenerative disease of old age whichmedical treatment may ameliorate, but not cure. Meanwhile, the expense of medicaltreatment, especially hospital treatment, has risen enormously and continues to rise, and it has been suggested that thecommunity is faced with a gloomy prospectof unlimited ‘medicated survival’ to be metby diminishing resources.From Nature 7 March 1953.

© 2003 Nature Publishing Group

usually involved in repairing damaged DNA.The same mechanism might be involved inthe transfer of neo-containing plastid DNA,although an earlier study8 of double-strand-break repair in tobacco found no evidence of plastid (or mitochondrial) DNA transferduring the process.

Given the readily detectable plastid-to-nucleus transfer of DNA, it is not surprisingthat the tobacco nuclear genome containslong tracts of plastid DNA. For example, asearch9 for plastid DNA fragments containinga particular gene, rbcL, found that there are 15 fragments of different size containing rbcLplastid DNA in the tobacco nuclear genome,the largest being 15.5 kilobases (kb) in size.But the amount of organellar DNA present in the nucleus seems to depend on the plantspecies concerned. For instance, the nucleargenome of Arabidopsis thaliana, the favouredmodel for plant biologists, contains only 17small insertions of plastid DNA (total size 11kb)10, and 13 small insertions (7 kb)10 and onelarge insertion (about 620 kb) of mitochondr-ial DNA11. The differences may in part be dueto different organelle-to-nucleus DNA trans-fer rates. Indeed, 40% of double-strand-breakrepair in the tobacco nucleus was accompa-nied by insertions (although, in the specificinstance, not of organellar DNA), whereas noinsertions accompanied double-strand-breakrepair in Arabidopsis8. If the conclusions of the study by Timmis et al.1 can be generalized,crops that have larger nuclear genomes arelikely to have higher transfer rates betweenplastids and the nucleus. But the relative abun-dance of organellar DNA in nuclear genomesmay also reflect the efficiency of mechanismsthat work against ‘genome obesity’ by selec-tively eliminating nuclear sequences12.

What about the biotechnological impli-cations of the new research? In general,incorporation of genes for some trait orother into the plastid genome is an attractivestrategy, as plastids in most crops are nottransmitted by pollen. So incorporation of,for instance, genes conferring herbicide orinsect resistance in the plastid genome wouldmean that these genes would not be trans-mitted by pollen to neighbouring stands ofthe same crop or to weedy relatives13. This is a common problem with transgenes incorporated in the nuclear genome14.

It is reassuring that the marker gene(aadA, which confers resistance to anotherantibiotic, spectinomycin) that Timmis and colleagues incorporated into plastidsalong with the kanamycin-resistance gene,was not expressed in the nucleus: this gene had a plastid-specific promoter. Butalthough aadA residing in the nucleus on a plastid genome segment is not expressed,broken plastid gene pieces, incorporatednext to a nuclear promoter, could beexpressed at a very low frequency15.

The authors tested for spectinomycinresistance only in the 16 plant lines carrying

the transferred neogene. To estimate the trans-fer rate of functional aadA, however, a seedlingpopulation of comparable size (250,000) orlarger would have to be screened for expres-sion of spectinomycin resistance. The fre-quency of fortuitous transfer that results inexpression of the transferred plastid gene maybe 100–100,000 times lower than the rate oftransfer of the nuclear neogene (the values willhave to be experimentally determined).

Nonetheless, incorporation of transgenesin plastids should still be effective for con-tainment of those genes. If a transgene isincorporated in the nucleus, each pollengrain will carry the transgenic trait. By con-trast, assuming that the probability of plastidgene expression from a broken fragment isonly 100 times lower than the transfer rate of plastid DNA, only 1 in 1.6 million pollengrains will carry the expressed plastid gene.

It could be, however, that functioning ofthe transferred plastid genes in the nucleuscould be prevented altogether. Examples ofgenes that are expressed in plastids but not inthe nucleus are bacterial genes containing ahigh content of adenine and thymine, such as those encoding the Bacillus thuringiensisinsecticidal protein or the Clostridium tetanitoxin13,16. The B. thuringiensis insecticidalgenes could be expressed in the plant nucleusonly by modifying them to a ‘eukaryoticdesign’17,18. So modifying a gene to a ‘bacterialdesign’ would be expected to facilitate itsexpression in plastids and prevent it in the nucleus.

Timmis and colleagues’ demonstration1

of the transfer of plastid DNA to the nucleusof higher plants will be the subject of muchdebate from both the scientific and biotech-nological viewpoints. The implications inthe latter respect will depend on the likeli-hood of plastid genes becoming functionalnuclear genes, and on the success of genedesigns that can prevent the expression oftransferred plastid genes in the nucleus. ■

Pal Maliga is at the Waksman Institute, Rutgers,The State University of New Jersey, Piscataway, New Jersey 08854-8020, USA.e-mail: maliga@waksman.rutgers.edu1. Huang, C. Y., Ayliffe, M. A. & Timmis, J. N. Nature 422, 72–76

(2003).

2. Abdallah, F., Salamini, F. & Leister, D. Trends Plant Sci. 5,

141–142 (2000).

3. Palmer, J. D. Nature 405, 32–33 (2000).

4. Sugiura, M. Plant Mol. Biol. 19, 149–168 (1992).

5. Thorsness, P. E. & Fox, T. D. Nature 346, 376–379 (1990).

6. Ricchetti, M., Fairhead, C. & Dujon, B. Nature 402, 96–100

(1999).

7. Yu, X. & Gabriel, A. Mol. Cell 4, 873–881 (1999).

8. Kirik, A., Salomon, S. & Puchta, H. EMBO J. 19, 5562–5566

(2000).

9. Ayliffe, A. M. & Timmis, J. N. Theor. Appl. Genet. 85, 229–238

(1992).

10.The Arabidopsis Genome Initiative. Nature 408, 796–815 (2000).

11.Stupar, R. M. et al. Proc. Natl Acad. Sci. USA 98, 5099–5103

(2001).

12.Bennetzen, J. L. Genetica 115, 29–36 (2002).

13.Maliga, P. Trends Biotechnol. 21, 20–28 (2003).

14.Riegel, M. A. et al. Science 296, 2386–2388 (2002).

15.Carrer, H., Hockenberry, T. N., Svab, Z. & Maliga, P.

Mol. Gen. Genet. 241, 49–56 (1993).

16.Tregoning, J. et al. Nucleic Acids Res. 31, 1174–1179 (2003).

17.Perlak, F. J., Fuchs, R. L., Dean, D. A., McPherson, S. L. &

Fischhoff, D. Proc. Natl Acad. Sci. USA 88, 3324–3328 (1991).

18.Murray, E. E. et al. Plant Mol. Biol. 16, 1035–1050 (1991).

news and views

32 NATURE | VOL 422 | 6 MARCH 2003 | www.nature.com/nature

Plastida

b

g1 g1

g2 g2

g3 g3

ptDNA

neo neo

aadA aadA

g1 g1

g2 g2

g3 g3

neo neo

aadA aadA

c

g1 g1

g2 g2

g3 g3

Nucleus

DNA

g1 g3neo aadA

g1 g3neo aadANo gene expression

Gene expression

Figure 1 Transfer of plastid DNA to the nucleusof a tobacco cell, as studied by Timmis andcolleagues1. a, Tobacco cell with a transformedplastid genome (ptDNA) and a wild-type, non-transformed nucleus. Three plastid genes(g1, g2, g3) are shown in the plastid genome,along with the kanamycin-resistance (neo) gene and the spectinomycin-resistance (aadA)gene. Crucially, neo has the necessary ancillarysequences for expression only in the nucleus,whereas aadA can be expressed only in theplastid. Only two ptDNA copies are shown out of the 1,000–10,000 copies present in a plant cell.b, A rare cell, in which a fragment of ptDNA hasbeen transferred to the nucleus: note that neo isexpressed here but aadA is not. c, Timmis et al.had to study the expression of plastid DNA in the nucleus in cells with wild-type ptDNA, the cells being created by pollinating non-transformed plants with pollen from plantspossessing transformed plastids. One in 16,000 pollen grains carried a neo gene insertedin a chromosome. Because plastids are nottransmitted by pollen, only the nuclear neogene (and flanking ptDNA) was transmitted inthe cross. Separation of plastid and nucleartransgene copies was essential as there are manymore plastid neo and aadA copies than nuclearcopies, and the large number (10,000 per leafcell) of plastid copies interferes with the study of the few copies in the nucleus.

© 2003 Nature Publishing Group